Abstract
Earthquake is considered as one of the biggest threats for the modern civilization. It ruins a large number of buildings throughout
the world. Many people die and huge amount of property is destroyed because of this dangerous natural disaster. This is not
possible to stop earthquake. So, the only way to save lives and properties from earthquake is the early prediction of it. Many
researches have been conducting on it and an expert system has been developed. But the developed system cannot handle
uncertainties which provides some inexact results. However, this paper presents a system that predicts earthquake from some
signs and patterns by using Evidential Reasoning Approach. The proposed system deals with uncertainties and produces the most
accurate results.
Key Words: Earthquake; natural disaster; Evidential Reasoning Approach; Signs and patterns.
Earthquake Prediction Using Evidential Reasoning Approach
1. IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
_______________________________________________________________________________________
Volume: 04 Issue: 12 | Dec-2015, Available @ http://www.ijret.org 149
EARTHQUAKE PREDICTION BY USING EVIDENTIAL REASONING
APPROACH
Subrina Akter1
, Muhammed Jamshed Alam Patwary2
, Shakhawat Hossain3
1
Lecturer, Department of CSE, International Islamic University Chittagong, Chittagong, Bangladesh
2
Assistant Professor, Department of CSE, International Islamic University Chittagong, Chittagong, Bangladesh
3
Computer Engineer, Chittagong, Bangladesh
Abstract
Earthquake is considered as one of the biggest threats for the modern civilization. It ruins a large number of buildings throughout
the world. Many people die and huge amount of property is destroyed because of this dangerous natural disaster. This is not
possible to stop earthquake. So, the only way to save lives and properties from earthquake is the early prediction of it. Many
researches have been conducting on it and an expert system has been developed. But the developed system cannot handle
uncertainties which provides some inexact results. However, this paper presents a system that predicts earthquake from some
signs and patterns by using Evidential Reasoning Approach. The proposed system deals with uncertainties and produces the most
accurate results.
Key Words: Earthquake; natural disaster; Evidential Reasoning Approach; Signs and patterns.
--------------------------------------------------------------------***----------------------------------------------------------------------
1. INTRODUCTION
Earthquake is one of the most audacious natural calamities.
An earthquake is defined as the shaking of ground because
of large sections of the earth‟s outermost crust .Earthquake
is considered as the most dangerous disaster which killed a
large number of people and destroyed a large amount of
properties. According to statistics 332723 earthquakes
occurred from 2000 to 2012 and 813856 people died in
these earthquakes [1]. Researchers say that only early
prediction of earthquake can minimize the number of death.
So, it is necessary to predict earthquake. This paper
represents a system that will predict earthquake from some
signs and patterns.
The paper is organized as follows: Section 1 provides the
introduction where section 2, 3, 4, 5 present background
study, literature review, proposed methodology and
prediction of earthquake, Section 6 represents the results and
discussion where section 7 concludes the paper.
2. BACKGROUND STUDY
As earthquake is considered the most dangerous natural
disaster, a number of researches have been conducted over
it. An expert system has also been developed to predict the
earthquake magnitude by Aojjatadell and panakkat[2]. But
this system is not capable of handling uncertainty which
provides inexact results.
However, this paper proposes a system that predicts
earthquakes from some signs and trends by using ER
approach. This system is capable of dealing with uncertainty
which provides the most correct results.
3. LITERATURE REVIEW
One of the most important branches of seismology is
earthquake prediction. Prediction of earthquake provides
some sufficient warning about the issue that verily happens
before earthquake [3].
Earthquake prediction can be the probabilistic assessment of
general earthquake branch. It is said that, earthquake
prediction is not yet precisely possible. But researches
argued that, there are some signs which predict earthquake,
for which they emphasis on some empirical analysis.
Identification of distinctive to earthquakes and geographical
trend or pattern in seismicity can make the prediction of
earthquake. An effective warning of earthquake can be done
by an earthquake precursor [4]. Precursor includes
animal‟s behavior, changes in VP/VS, random
emissions and electromagnetic variation. The unusual
behavior of animal are referred to as the response to p-wave
[5]. This is very often observed before the earthquake.
Changes in VP/VS are an important issue of earthquake
prediction. VP denotes the velocity of seismic “p”
(primary/pressure) wave passing through the rock and VS is
the symbol for the velocity of the “s”(secondary/shear)
wave. Some earthquake laboratory examination show that
earthquake occurs when the ratio of VP and VS changes –
from the normal values [6]. Radon is one of the most-
important gases that are contained by the most of rocks. The
random emission of Radon occurs before earthquake.
Another important issue of precursors is the variation in
electro-resistive or magnetic phenomenon.
Another approach of predicting earthquakes is predicting
earthquake by looking for trends or pattern that leads to an
earthquake which include elastic rebound, seismic gap and
2. IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
_______________________________________________________________________________________
Volume: 04 Issue: 12 | Dec-2015, Available @ http://www.ijret.org 150
seismicity patterns. Earthquake may occur due to fluid
intrusion and stress triggering which is known as seismicity
patterns.
4. OVERVIEW OF EVIDENTIAL REASONING
Evidential Reasoning Approach is developed based on
Dempster Shafer Theory. This approach works on Multiple
Attribute Decision Analysis (MADA). In this approach, the
basic attributes of Architectural Theory Diagram (ATD) are
aggregated. For that, weights and belief are assigned for
each basic attribute by the domain expert. The mass of each
attribute are calculated by using the following formula.
mj,k = wk ∗ βij , j = 1, … . . . , N (1)
and
mD,k= j,k (2)
Here, mD,k is the probability mass unassigned to any
individual consequent. This mD,k is decomposed into ̅mD,k
and m˜ D,k [7] where ̅mD,k = 1 – wk (3)
and m˜ D,k = wk ∗ j,k (4)
Then the overall aggregated degree of belief βj and Dj is
generated as:
{Dj}:mj,I(k+1) = KI(k+1)[mj,I(k)mj,k+1 +mj,I(k)mD,k+1 +
mD,I(k)mj,k+1] (5)
mD,I(k) = mD,I(k) + m˜ D,I(k) k = 1, … … , L
{D}:m˜ D,I(k+1) = KI(k+1)[m˜ D,I(k)m˜ D,k+1 +
m˜ D,I(k)mD,k+1 + mD,I(k)m˜ D,k+1] (6)
{D}:mD,I(k+1)= KI(k+1)[mD,I(k)mD,k+1] (7)
= -1
(8)
Here, k=1………L-1.
(9)
(10)
Here, βD is the remaining belief degree unassigned to any
Dj.
5. PREDICTION OF EARTHQUAKE
Prediction of earthquake is done by using ER approach .For
that, the user provides the inputs for some linguistic term
say high, medium or low. The results of the system are
shown in percentage. The unknown results are also shows in
this system.
(Presentation) (Application) (Data Processing)
Fig-1: System Architecture to predict Earthquake
Here, presentation layer takes input and Provides output.
Application layer consist of application logic and
prediction method. Data process layer consist of database
and prediction parameters.
Fig-2: System framework
The framework of this system is shown in figure 2 as an
ATD. This prediction is done by using formula from (1) to
(9). The unknown results are calculated by using formula
(10).
6. RESULTS AND DISCUSSION
The proposed system aggregates the values provided by the
user for each attributes. These values are the belief of the
user. User needs to provide inputs for high, medium and low
for each attribute. The total input will be 1 or less than 1 for
each attribute.
Confidencelevel
43.5%
Medium LowHigh
23%
30%
Fig-3: Earthquake prediction results
3. IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
_______________________________________________________________________________________
Volume: 04 Issue: 12 | Dec-2015, Available @ http://www.ijret.org 151
The system shows the output for high, medium and low in
percentages. The total output will be 100% or less than
100%. If the results are less than 100%, then the remaining
values will be the unknown values which are actually the
uncertainty that the system handles. Manual system provides
the inexact results as it cannot handle uncertainties. A ROC
(receiver operating characteristics) curve is used to show
that the proposed system performs better than the manual
system.
Fig-4: Performance comparison between the manual
system and proposed system by ROC curve
Here, the green curve shows the expert system‟s
earthquake prediction performance and the blue curve
shows the prediction performance of the manual system.
The AUC (area under curve) of the green curve is 0.975
where the AUC of blue curve is 0.850. So, it becomes
clear that, the proposed system is more efficient in
earthquake prediction than the manual system.
7. CONCLUSION
ER approaches is used to predict earthquakes from some
signs and patterns. ER approached can handle uncertainties
which are the big limitation of the existing systems.
Prediction of earthquake is possible if the signs and patterns
can be observed in time. But this not necessary to observe
all the signs and patterns to predict earthquake. If any sign
or pattern is not seen then the user may provide a minimum
value say 0% or 0.1% for that sign or pattern.
In future, the system will be developed by using Rule Base
Interface Methodology using Evidential Reasoning
(RIMER) Approach. Then, the system will be more reliable
and provide more accurate results.
REFERENCES
[1].earthquake.usgs.gov/earthquakes/eqarchives/year/sqststs.
php
[2]. Hojjat Adeli, Ashif Panakkat „A probabilistic neural
network for earthquake magnitude prediction, Neural
Networks 22 (2009) 1018–1024
[3]. Geller, Robert J.; Jackson, David D.; Kagan, Yan Y.;
Mulargia, Francesco (14 March 1997), "Earthquakes Cannot
Be Predicted", Science 275 (5306): 1616,
doi:10.1126/science.275.5306.1616.
[4]. Geller,RobertJ(December1997),"Earthquake
prediction: a critical review." (PDF), Geophysical
Journal International 131 (3):425–450,
Bibcode:1997GeoJI.131..425G,doi:10.1111/j.1365-
246X.1997.tb06588.x
[5]. International Commission on Earthquake Forecasting
for Civil Protection (30 May 2011), "Operational
Earthquake Forecasting: State of Knowledge and Guidelines
for Utilization", Annals of Geophysics 54 (4): 315–391,
doi: 10.4401/ag-5350 (inactive 2015-01-12)
[6]. Hammond, Allen L. (23 May 1973), "Earthquake
predictions: Breakthrough in theoretical insight?", Science
180(4088):851853,Bibcode:1973Sci...180.851H,doi:10.1126
/science.180.4088.851, PMID 17789254.
[7]. Xinlian Xie • Dong-Ling Xu • Jian-Bo Yang•Jin Wang
Jun Ren • Shijun Yu; Ship selection using a multiple-criteria
synthesis approach; J Mar Sci Technol (2008) 13:50– 62,
DOI 10.1007/s00773-007-0259-4
BIOGRAPHIES
Subrina Akter, MS(Engg.) and B. Sc
Honors in CSE from Chittagong
University (CU). Lecturer in the
Department of CSE at International
Islamic University Chittagong,
Bangladesh. Research interests include
semantic web, machine learning and
data mining.
Muhammed Jamshed Alam Patwary
received the B. Sc and MS(Engg.)
degree in Computer Science and
Engineering from University of
Chittagong in 2006 and 2008
respectively. Assistant Professor in the
department of Computer Science and
Engineering at International Islamic
University Chittagong (IIUC). His major research interests
include decision support system, expert system, machine
learning and data mining.
Shakahwat Hossain received the B. Sc
(Engg.) degree in Computer Science
and Engineering from International
Islamic University Chittagong (IIUC).
His major research interests include
decision support system, expert system,
machine learning and data mining.